INDEX

Explanations

phrases that indicate novelty or innovation

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

IFF

-0.63

 Bastard

-0.61

MPG

-0.61

 Brach

-0.60

 PRESS

-0.60

OPA

-0.59

 Donation

-0.59

iff

-0.57

 Originally

-0.57

iano

-0.56

POSITIVE LOGITS

bies

1.31

bie

1.27

 Zealand

1.03

 paradigm

0.88

foundland

0.85

 millennium

0.85

 territory

0.82

rums

0.82

 dimension

0.81

found

0.80

Activations Density 0.094%

phrases that indicate novelty or innovation

No Comments

No Known Activations